AN APPLIED RESEARCH AIMING TO IM-PROVE ORGANISATION’S DATA QUALITY: A RULE-BASED CLASSIFICATION FREAMEWORK FOR DETECTING BAD DA-TA MATCHING

Abstract

Data quality continues being a problem for organizations. Data quality problems occurs not only among single data collections, but also multiple data source especially when dealing with data integrity. The paper proposed a framework to re-evaluate the data matching result that how it was done is unknown. It aims at improving data quality and preventing wrong fed-in data. The main focus is the choice of classification methods. In this study, rule-based classification, decision tree and k-means clustering was considered as applicable choices. Thus need assessments and performance evaluation was conducted to exam if candidate approaches meet the subjective perceptions of stakeholders and objective task requirements.